Re: suggestion: log_statement = sample - Mailing list pgsql-general

From Janning Vygen
Subject Re: suggestion: log_statement = sample
Date
Msg-id 200907211801.38976.vygen@kicktipp.de
Whole thread Raw
In response to Re: suggestion: log_statement = sample  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: suggestion: log_statement = sample
List pgsql-general
On Tuesday 21 July 2009 15:49:36 Tom Lane wrote:
> Janning Vygen <vygen@kicktipp.de> writes:
> > On Monday 20 July 2009 19:24:13 Bill Moran wrote:
> >> Have you benchmarked the load it creates under your workload?
> >
> > Yes, it takes up to 15% of our workload in an average use case. But we
> > have peak times where we can not afford 15% lost for logging!
>
> Well, you could turn it off during the peak times.

It affords a server restart which is not a good idea. And i will never get any
real time analyze of what happens at peak time.

> Or buy better
> hardware --- if you have less than 15% headroom then it's past time
> to be doing that anyway.

yes, you are right, but software solutions are so much nicer...

> You really haven't given any convincing case for this behavior...

that's fine more me. I still don't understand it why it was not convincing. I
guess you guys run your DB with lots of hardware power, RAIDs and NAS and so
on. But there are many people running it on small boxes with one IDE hard
disk. Switching full logging on with such hardware is like shutting it down as
IO is blocked! I still think it has reasonable uses cases without lots of
effort (just a guess, i havn't looked at the source code yet).

But, hey: I just wanted to help improving postgresql. If you think this is not
an improvment, its fine for me. Thanks to all for discussing it.

kind regards
Janning


pgsql-general by date:

Previous
From: David Kerr
Date:
Subject: Re: killing processes
Next
From: Tom Lane
Date:
Subject: Re: suggestion: log_statement = sample